Published 2026-09-21 · Tested 2026-08-20
n8n
An8n received 10 PASS votes and passed 5 of five agent surface checks. The clearest finding came from the find the exact limits task.
Three AI models, GPT 5.6 Sol, Claude Opus 5, and DeepSeek v4 Flash, each read n8n’s public documentation independently and attempted five first-hour developer jobs: build the first workflow, find the exact limits, recover a node from a 429, secure an incoming webhook, paginate workflows.
No accounts, API calls, or code execution were used. Every verdict came from public pages and every published quotation passed a live verification check. Preflight substitutions: T5: Public API pagination and filtering replace SDK coverage.
Freshness
How rechecks work- Category
- Workflows & automation
- Tested
- Quotes verified
- Surface rechecked
No change since the test.
5 of 5 quoted passages still appear on the live pages.
93.8% · 75/80 · AI Agent Readiness Score · reading 30 pts · surface 50 pts
Agent surface checks · 50/50
The Reading Test
| Task | GPT 5.6 Sol | Opus 5 | DeepSeek v4F | Consensus |
|---|---|---|---|---|
| Build the first workflow | PASS | PASS | PASS | PASS |
| Find the exact limits | PARTIAL | PARTIAL | PARTIAL | PARTIAL |
| Recover a node from a 429 | PASS | PASS | PASS | PASS |
| Secure an incoming webhook | PASS | PASS | PASS | PASS |
| Paginate workflows | PARTIAL | PARTIAL | PASS | PARTIAL |
docs platform: GitBook (unscored) · verified 2026-08-20
What to fix first
These 2 fixes could add up to 5 points to the AI Agent Readiness Score. The list ranks each fix by the points it would add. How the ranking works
- 1
Found: The docs omit per-plan execution and concurrency limits, and pages use different plan names. GPT also found Pro retention of 30 days in the docs but 14 on support.n8n.io.
Fix: Add each Cloud plan's execution and concurrency limits to the Manage your data page. Use the same plan names and Pro retention value on every page.
Evidence: docs.n8n.io/
deploy/ use-n8n-cloud/ configure-cloud/ manage-your-data (opens in a new tab) - 2
Found: The Pagination page's next-page curl examples omit the X-N8N-API-KEY header that its first-page examples include. The page also does not say when cursor paging ends.
Fix: Add the X-N8N-API-KEY header to every next-page curl example on the Pagination page. State that paging ends when nextCursor is null.
Evidence: docs.n8n.io/
connect/ n8n-api/ pagination (opens in a new tab)
What the docs get right
- Build the first workflow: 3 PASS votes. One linear quickstart covers the schedule, NASA credential, date expression, manual execution, output inspection, Postbin check, and publishing.
- Recover a node from a 429: 3 PASS votes. The rate-limit guide identifies 429, configures retry delay, and shows the Loop Over Items-API-Wait cycle.
- 5 of 5 agent surface checks. Present: llms.txt, llms-full.txt, markdown mirrors, an MCP server, docs AI.
Build the first workflow
PASSPASS consensus from 3 PASS.
One linear quickstart covers the schedule, NASA credential, date expression, manual execution, output inspection, Postbin check, and publishing. One numbered path covers create, Schedule Trigger settings, NASA credential creation, the literal expression {{ $today.minus(7, 'days') }}, per-node Execute step, full Execute Workflow, and Publish, with no gaps requiring a guess. Step-by-step tutorial covers Schedule Trigger, NASA node with DONKI solar flare, credential creation from NASA API, expression for date calculation, If node with classType, PostBin output, and publishing. Path is unambiguous.
Find the exact limits
PARTIALPARTIAL consensus from 3 PARTIAL.
The values require Docs, Pricing, and Support, and Pro retention conflicts at 30 days versus 14 days. Saved-execution counts, retention windows, and the 100GB storage cap are stated precisely on the Manage your data page, but no per-plan concurrent-execution number appears anywhere in the docs and per-plan execution allowances are only implied by the memory-tier labels Pro-1 (10k executions) and Pro-2 (50k executions). Execution allowances per plan are not documented in the docs. Saved-execution counts and retention windows are documented on the manage-your-data page (Start/Starter: 2500/7 days, Pro: 25000/30 days, Enterprise: 50000/unlimited). Concurrent-execution limits appear only on the external pricing page (5/20/200+). Plan naming is inconsistent across pages: 'Start and Starter' vs 'Pro-1 (10k executions)' vs 'Pro-2 (50k executions)' vs 'Pro'. Values require cross-referencing three pages and the external pricing site.
Recover a node from a 429
PASSPASS consensus from 3 PASS.
The rate-limit guide identifies 429, configures retry delay, and shows the Loop Over Items-API-Wait cycle. A single page names the 429 symptom, then gives complete click-paths for both remedies including where to connect the Wait node back to Loop Over Items and how to size Wait Between Tries (ms) against the service's rate. Two complete documented strategies: (1) Retry On Fail in node Settings with Max Tries and Wait Between Tries, and (2) Loop Over Items node with Wait node. Both are clearly explained with configuration steps. The HTTP Request node common-issues page also documents both Batching and Retry on Fail for the 429 case.
Secure an incoming webhook
PASSPASS consensus from 3 PASS.
The Webhook and credential pages cover test listening, Header/JWT credentials, publishing, and the production URL. The node page, the Workflow development page, and the Webhook credentials page together cover the test URL and its 120-second Listen for test event window, the Basic, Header, and JWT credential fields, publishing, and the switch to the production URL. Complete documentation: test URL with Listen for test event, production URL after publishing, four authentication methods (None, Basic auth, Header auth, JWT auth), IP allowlist, and the workflow-development page covering the build-test-publish cycle. JWT credential types are documented separately (passphrase and PEM key).
Paginate workflows
PARTIALPARTIAL consensus from 1 PASS, 2 PARTIAL.
The next-page examples omit X-N8N-API-KEY and do not state the stop condition for cursor traversal. Authentication, the active=true filter, the cursor parameter, and the nullable nextCursor termination signal are all documented, but the Pagination page's next-page curl examples omit the X-N8N-API-KEY header that its own first-page examples include, so copying them verbatim fails authentication. Full documentation: API key creation under Settings > n8n API, authentication via X-N8N-API-KEY header, filtering with ?active=true, cursor-based pagination with nextCursor, default page size 100 (max 250), and a complete curl example showing the request and response pattern. The workflow endpoint reference also documents the active parameter and cursor field.
The receipt
Pro plans: max 25000 executions saved and 30 days execution log retention;
The values require Docs, Pricing, and Support, and Pro retention conflicts at 30 days versus 14 days.
Agent surface notes
Initialize returned a valid MCP protocol response.
The live docs expose the public GitBook AI Answer assistant.
Show the score
Paste this into a readme:
[](https://docsforagents.com/reports/n8n-docs-ai-agent-readiness/) Method note
This is a reading test of public documentation, not an execution test. No accounts were created and no API calls were run. The AI Agent Readiness Score counts fifteen reading votes at PASS 2, PARTIAL 1, and FAIL 0, for 30 possible points. Five agent surface checks add 10 points each. The total is 80. Consensus chips show each row majority and do not affect scoring. The panel split on 1 of five tasks. Quotes shown here were re-fetched and confirmed verbatim on 2026-08-20.
Put another docs site through the battery.
Nominate a docs site